Adverse Perinatal Outcomes Associated With Timing of Initiation of Antiretroviral Therapy

Systematic Review and Meta-Analysis

Harriet Sexton; Mary Kumarendran; Zoe Brandon; Christine Shi; Shona Kirtley; Joris Hemelaar

HIV Medicine. 2023;24(2):111-129.

Abstract and Introduction

Abstract

Background: The World Health Organization (WHO) recommends immediate initiation of lifelong antiretroviral therapy (ART) for all people living with HIV, including pregnant women. As a result, an increasing number of women living with HIV conceive while taking ART, the vast majority of whom reside in low- and middle-income countries (LMICs). We aimed to assess the association between timing of ART initiation and perinatal outcomes.

Methods: We conducted a systematic literature review by searching PubMed, CINAHL (EBSCOhost), Global Health (Ovid), EMBASE (Ovid), and the Cochrane Central Register of Controlled Trials and four clinical trial databases (WHO International Clinical Trials Registry Platform, the Pan African Clinical Trials Registry, the ClinicalTrials.gov database, and the ISRCTN Registry) from 1 January 1980 to 28 April 2018. We identified studies reporting specific perinatal outcomes among pregnant women living with HIV according to timing of ART initiation and extracted data. Perinatal outcomes assessed were preterm birth (<37 weeks), very preterm birth (<32 weeks), low birthweight (<2500 g), very low birthweight (<1500 g), small for gestational age (<10^th centile), very small for gestational age (<3^rd centile) and neonatal death (<29 days). Random-effects meta-analyses examined perinatal outcomes associated with preconception and antenatal ART initiation as well as according to trimesters of antenatal initiation. We performed quality assessments and subgroup and sensitivity analyses, and assessed the effect of adjustment for confounders. This systematic review and meta-analyses is registered with PROSPERO, number CRD42021248987.

Results: Of 51 874 unique citations, 25 studies (eight prospective and 17 retrospective cohort studies) were eligible for analysis, including 40 920 women living with HIV. Preconception ART initiation was associated with a significantly increased risk of preterm birth (relative risk 1.16; 95% confidence interval [CI] 1.03–1.31) compared with antenatal ART initiation. Preconception ART initiation was not significantly associated with very preterm birth, low birthweight, very low birthweight, small for gestational age, very small for gestational age, or neonatal death. First trimester exposure (i.e. preconception or first trimester initiation) was not significantly associated with any increased risk of adverse perinatal outcomes. No significant association between timing of ART initiation and adverse perinatal outcomes was found in the studies of higher quality and those conducted in LMICs.

Conclusion: Preconception ART initiation is associated with preterm birth but no other adverse perinatal outcomes. In LMICs, where most pregnant women living with HIV reside, the timing of ART initiation was not associated with any adverse perinatal outcomes.

Introduction

Of the 38.0 million people currently living with HIV, 15.5 million are women aged 15–49 years, of whom 1.3 million are pregnant each year.^[1] The vast majority of pregnant women living with HIV reside in low- and middle-income countries (LMICs). In 2013, the World Health Organization (WHO) recommended that all pregnant women living with HIV should receive antiretroviral therapy (ART) to prevent vertical transmission of HIV.^[2] This led to an increase in the global proportion of pregnant women living with HIV who received ART during pregnancy, from 44% in 2010 to 82% in 2018, resulting in a 41% reduction in vertical HIV transmission in the same period.^[1] In 2015, the WHO adopted a 'treat all' approach, which recommended that all people living with HIV, including pregnant women, should initiate lifelong ART as soon as possible after diagnosis.^[3] This resulted in a dramatic increase in the proportion of pregnant women living with HIV who received ART at the time of conception, from 7% in 2010 to 51% in 2018, in the 23 focus countries in which 86% of global pregnant women living with HIV reside.^[4] Preconception ART has been linked to extremely low rates of vertical transmission of HIV^[5,6] while also improving maternal health. Despite this, ART during pregnancy has potential adverse effects.

LMICs carry the highest burden of neonatal and child mortality.^[7] Preterm birth is the most important cause of neonatal and child mortality globally,^[8] with an estimated 14.8 million babies born preterm each year.^[9] Furthermore, 23.3 million small for gestational age babies born each year account for 21.9% of neonatal deaths in LMICs.^[10] Preterm birth and small for gestational age are both causes of low birthweight (18 million/year), an outcome measure frequently used in LMICs as gestational age at birth is often unknown.^[11] United Nations Sustainable Development Goal 3 target 3.2 aims to reduce preventable deaths of newborns and children aged <5 years,^[12] a goal that will be difficult to achieve without improving the perinatal outcomes that are the basis of many of these deaths.

Pregnant ART-naïve women living with HIV are associated with increased risk of preterm birth (relative risk [RR] 1.50; 95% confidence interval [CI] 1.24–1.82), low birthweight (RR 1.62; 95% CI 1.41–1.86), small for gestational age (RR 1.31; 95% CI 1.14–1.51), and stillbirth (RR 1.67; 95% CI 1.05–2.66).^[13] ART in pregnancy has also been linked to increased rates of adverse perinatal outcomes, although studies have shown conflicting results related to regimen complexity (e.g. zidovudine monotherapy vs. ART, i.e. triple drug therapy) and type of regimen/class of drugs (e.g. protease inhibitors [PIs]).^[14,15] Some studies have shown an increased risk of preterm birth with antenatal initiation of ART compared with zidovudine monotherapy,^[16,17] but this was not seen in other studies.^[18,19] Similarly, PI-based ART was associated with an increased risk of preterm birth in a number of studies^[20,21] but not in others.^[22] A recent network meta-analysis of randomized controlled trials (RCTs) of ART initiated during pregnancy found that PI lopinavir/ritonavir-containing regimens were associated with the highest risks of adverse perinatal outcomes.^[23]

A key outstanding question is whether the timing of ART initiation impacts perinatal outcomes in pregnant women living with HIV. We conducted a systematic review and meta-analysis of studies reporting the association of adverse perinatal outcomes with preconception and antenatal ART initiation as well as according to trimesters of antenatal initiation.

1 2 3 4

Next Section

Abstract and Introduction
Methods
Results
Discussion
References

Table 1. Characteristics of studies included in the systematic review and meta-analysis
Study	Country	Country income	Recruitment period	Number of women analysed	Population characteristics	Methods to correct for confounders	Gestational age estimation method	Quality assessment
Prospective cohorts
Aaron (2012)[30]	USA	High	1/2000 to 1/2011	183	First-born twin included, 38.3% smoking, 18.0% IDU, urban setting	Regression analysis	LNMP confirmed by second trimester ultrasound	Average
Ekouevi (2008)[34]	Ivory Coast	Low/middle	3/2001 to 8/2007	139	Twins excluded, recruited from antenatal clinics, urban setting	None	Symphysis fundal height	Poor
Kowalska (2003)[36]	Poland	High	1/1995 to 2/2003	40	Twins included, 47.1% IDU, recruited from an outpatient HIV clinic	None	LNMP	Poor
Li (2016)[18]	Tanzania	Low/middle	11/2004 to 10/2011	1094	Recruited from hospitals, health centres, and dispensaries, urban setting	None	LNMP	Poor
Machado (2009)[38]	Brazil	Low/middle	1996 to 2006	304	Twins excluded, 5.4% alcohol use, 21.3% smoking, 9.0% IDU, recruited from an HIV reference centre, urban setting	Regression analysis and risk factor analysis	LNMP or ultrasound	Poor
Malaba (2017)[39]	South Africa	Low/middle	4/2013 to 8/2015	1554	Twins excluded, recruited from large community primary care facility, urban setting	Regression analysis	LNMP	Average
Rudin 2011[44]	Switzerland	High	1984 to 2007	418	Twins excluded, 22% smoking, 26% IDU	None	No description	Poor
Townsend 2007[49]	United Kingom and Ireland	High	1990 to 2005	2201	Singleton pregnancies, 5.0% IDU	Regression analysis	No description	Poor
Retrospective cohorts
Aniji (2013)[31]	South Africa	Low/middle	1/10/2007 to 31/3/2009	245	Twins excluded, alcohol, smoking and IDU was negligible, all tertiary hospital deliveries	None	Late ultrasound	Poor
Chagomerana (2017)[32]	Malawi	Low/middle	1/4/2012 to 15/11/2015	2909	Twins excluded, all women delivered in a secondary hospital	Regression analysis	LNMP	Average
Chen (2012)[16]	Botswana	Low/middle	1/5/2009 to 30/4/2011	3081	First-born twin included, 5.3% alcohol use, 1.7% smoking, all hospital deliveries	Regression analysis	LNMP, symphysis-fundal height or ultrasound	Average
Chetty (2018)[33]	South Africa	Low/middle	2010 to 2015	2021	Alcohol, smoking and IDU under 10%. Enrolment clinic: 56.6% rural, 41.9% preurban	None	LNMP, symphysis-fundal height or ultrasound	Poor
Favarato (2018)[35]	United KIngdom and Ireland	High	2007 to 2015	6073	Twins excluded, 1.7% IDU	Regression analysis	No description	Poor
Lopez (2014)[37]	Spain	High	1/2006 to 12/2011	149	Twins excluded, 31.4% smoking, 6.4% IDU, recruited in a tertiary hospital, urban setting	Risk factor analysis	First trimester ultrasound	Average
Mandelbrot (2015)[6]	France	High	2000 to 2011	8678	Twins included, women recruited from 90 perinatal centres, all hospital deliveries	None	LNMP and/or ultrasound	Poor
Montgomery-Taylor (2015)[40]	United Kingdom	High	1/2008 to 12/2012	61	13.0% alcohol use, 3.0% smoking, recruited in a tertiary hospital, urban setting, all hospital deliveries	None	No description	Poor
Oomeer (2015)[41]	United Kingdom	High	2007 to 2014	37	Twins included	None	No description	Poor
Ramokolo 2017[42]	South Africa	Low/middle	10/2012 to 5/2013	1396	Recruited from primary health facilities	Regression analysis	LNMP	Poor
Rempis (2017)[43]	Uganda	Low/middle	2/2013 to 12/2013	94	Twins excluded, all delivered in a private referral hospital	None	No description	Poor
Samuel (2014)[45]	United Kingdom	High	1/3/2004 to 1/12/2010	99	Twins excluded, 3.4% IDU, recruited from HIV specialist centres, urban setting	None	No description	Poor
Sebitloane (2017)[46]	South Africa	Low/middle	1/4/2011 to 30/4/2014	725	Twins excluded, women recruited and delivered at a regional hospital, urban setting	None	No description	Poor
Short (2014)[47]	United Kingdom	High	1999 to 2010	251	Twins included, 13.0% smoking, recruited from an HIV antenatal clinic, all deliveries in a tertiary hospital, urban setting	None	No description	Poor
Snijdewind (2018)[48]	Netherlands	High	1/1997 to 2/2015	1392	Twins included, 6.1% alcohol use, 8.2% smoking, 1.7% IDU	Risk factor analysis	LNMP or early ultrasound	Average
Zash (2016)[51]	Botswana	Low/middle	5/2009 to 4/2011 and 4/2013 to 4/2014	4392	Twins included, recruitment and delivery at the two largest public maternity wards in the country, all hospital deliveries	None	LNMP	Poor
Zash (2018)[50]	Botswana	Low/middle	3/2013 to 8/2016	3384	Twins excluded, 8.8% alcohol consumption or smoking	Regression analysis	LNMP confirmed by ultrasound where possible	Average

Abbreviations: IDU, intravenous drug user; LNMP, last normal menstrual period.

Table 2. Antiretroviral therapies, timing of ART initiation and perinatal outcomes reported by studies included in the meta-analysis
Study	ART	Preconception versus antenatal	Preconception/T1 versus T2/T3	T1 versus T2	T2 versus T3	T1 versus T3	Outcome
Prospective cohorts
Aaron (2012)[30]	63.9% PI-based ART 21.3% NNRTI-based ART 14.8% NRTI-based ART	Yes	Yes	Yes	Yes	Yes	Small for gestational age, very small for gestational age
Ekouevi (2008)[34]	100% NNRTI-based ART	Yes	No	No	No	No	Low birthweight
Kowalska (2003)[36]	60.9% non-PI-based ART 39.1% PI-based ART	Yes	Yes	Yes	Yes	Yes	Preterm birth
Li (2016)[18]	94.0% NNRTI-based ART 4.5% not reported 1.5% PI-based ART	Yes	No	No	No	No	Preterm birth, low birthweight, small for gestational age, very small for gestational age
Machado (2009)[38]	68.1% PI-based ART 31.9% NNRTI-based ART	Yes	No	No	No	No	Preterm birth, low birthweight
Malaba (2017)[39]	97.4% NNRTI-based ART 2.6% PI-based ART	Yes	Yes	Yes	Yes	Yes	Preterm birth, very preterm birth, low birthweight, very low birthweight, small for gestational age
Rudin 2011[44]	84.0% PI-based ART 16.0% non-PI-based ART	Yes	No	No	No	No	Preterm birth
Townsend 2007[49]	54.1% NNRTI-based ART 37.1% PI-based ART 6.7% ART with PI + NNRTI 2% NRTI-based ART	No	Yes	No	No	No	Preterm birth
Retrospective cohorts
Aniji (2013)[31]	95.5% NNRTI-based ART 4.5% PI-based ART	Yes	No	No	No	No	Low birthweight, very low birthweight
Chagomerana (2017)[32]	100% NNRTI-based ART	Yes	Yes	Yes	No	No	Preterm birth, very preterm birth
Chen (2012)[16]	90.1% NNRTI-based ART 9.9% PI-based ART	Yes	No	No	No	No	Preterm birth, small for gestational age, neonatal death
Chetty (2018)[33]	100% NNRTI-based ART	Yes	No	No	No	No	Preterm birth, very preterm birth, low birthweight, small for gestational age
Favarato (2018)[35]	69.0% PI-based ART 31.0% NNRTI-based ART	Yes	No	No	No	No	Preterm birth
Lopez (2014)[37]	67.9% PI-based ART 32.1% NNRTI-based ART	Yes	No	No	No	No	Small for gestational age
Mandelbrot (2015)[6]	76.0% PI-based ART 15.8% NNRTI-based ART 5.9% NRTI-based ART	Yes	Yes	Yes	Yes	Yes	Preterm birth, very preterm birth
Montgomery-Taylor (2015)[40]	55.9% PI-based ART 37.2% NNRTI-based ART 6.9% other ART	Yes	No	No	No	No	Small for gestational age
Oomeer (2015)[41]	100% PI-based ART	Yes	No	No	No	No	Low birthweight
Ramokolo 2017[42]	100% NNRTI-based ART	Yes	No	No	No	No	Preterm birth, low birthweight, small for gestational age
Rempis (2017)[43]	100% NNRTI-based ART	Yes	No	No	No	No	Small for gestational age
Samuel (2014)[45]	100% PI-based ART	Yes	No	No	No	No	Preterm birth, low birthweight
Sebitloane (2017)[46]	100% NNRTI-based ART	Yes	No	No	No	No	Preterm birth
Short (2014)[47]	57.6% NNRTI-based ART 40.3% PI-based ART 2.1% NRTI-based ART	Yes	No	No	No	No	Preterm birth
Snijdewind (2018)[48]	66.7% PI-based ART 31.5% NNRTI-based ART 0.9% PI+NNRTI or NRTI-based ART	Yes	No	No	No	No	Preterm birth, very preterm birth, low birthweight, very low birthweight, small for gestational age
Zash (2016)[51]	37.0% TDF/FTC/EFV 63.0% other ART	Yes	No	No	No	No	Preterm birth, small for gestational age
Zash (2018)[50]	100% NNRTI-based ART	Yes	Yes	Yes	No	No	Preterm birth, very preterm birth, low birthweight, very low birthweight, small for gestational age, very small for gestational age, neonatal death

Notes: ARTs are given for each study as a whole. Detailed information on ART regimens according to timing of ART initiation is presented in Appendix 3.1. If ART regimens differed by timing of ART initiation arm in individual studies, then the proportions of ART regimens in each arm were weighed according to the number of people in each arm. Abbreviations: ART, antiretroviral therapy; EFV, efavirenz; FTC, emtricitabine; NNRTI, non-nucleoside reverse transcriptase inhibitor; NRTI, nucleoside reverse transcriptase inhibitor; PI, protease inhibitor; TDF, tenofovir; T1, first trimester; T2, second trimester; T3, third trimester.

Table 3. Random-effects meta-analysis results for relative risk of perinatal outcomes for different comparisons of timing of ART initiation. (a) All studies. (b) Studies conducted in high-income countries. (c) Studies conducted in low- and middle-income countries
Timing of ART initiation	Perinatal outcome
Timing of ART initiation	Preterm birth	Very preterm birth	Low birthweight	Very low birthweight	Small for gestational age	Very small for gestational age	Neonatal death
(a) All studies
Preconception vs antenatal	1.16 (1.03–1.31)	1.16 (0.90–1.49)	1.08 (0.91–1.28)	1.12 (0.76–1.66)	1.04 (0.83–1.30)	0.89 (0.60–1.32)	0.93 (0.59–1.47)
Preconception/T1 vs T2/T3	1.08 (0.95–1.24)	1.18 (0.97–1.43)	1.06 (0.88–1.28)	0.98 (0.71–1.33)	0.96 (0.82–1.12)	0.71 (0.38–1.30)	1.24 (0.75–2.06)
T1 vs T2	1.09 (0.99–1.21)	1.04 (0.81–1.33)	1.17 (0.79–1.71)	1.03 (0.71–1.49)	1.09 (0.80–1.49)	0.86 (0.62–1.19)	1.53 (0.87–2.68)
T2 vs T3	0.97 (0.61–1.54)	5.64 (2.70–11.78)	0.88 (0.57–1.37)	8.66 (0.51–146.11)	0.81 (0.56–1.18)	3.97 (0.56–28.16)
T1 vs T3	1.19 (0.91–1.55)	5.39 (2.41–12.09)	1.32 (0.80–2.17)	10.97 (0.60–202.14)	1.01 (0.64–1.59)	2.42 (0.27–21.63)
(b) High-income countries
Preconception vs antenatal	1.07 (0.87–1.31)	1.47 (1.19–1.82)	0.60 (0.17–2.16)	1.75 (0.98–3.11)	0.92 (0.63–1.36)	0.28 (0.07–1.16)
Preconception/T1 vs T2/T3	1.21 (0.96–1.51)	1.41 (1.11–1.78)			0.71 (0.44–1.14)	0.43 (0.17–1.10)
T1 vs T2	1.98 (0.33–12.02)	0.90 (0.58–1.41)			0.76 (0.38–1.51)	0.61 (0.19–1.92)
T2 vs T3	1.08 (0.46–2.53)	5.44 (2.54–11.66)			1.07 (0.55–2.08)	3.97 (0.56–28.16)
T1 vs T3	1.32 (1.05–1.66)	4.92 (2.12–11.40)			0.81 (0.34–1.94)	2.42 (0.27–21.63)
(c) Low- and middle-income countries
Preconception vs antenatal	1.15 (0.97–1.36)	0.95 (0.74–1.21)	1.07 (0.91–1.25)	0.89 (0.60–1.33)	1.08 (0.82–1.43)	1.00 (0.79–1.26)	0.93 (0.59–1.47)
Preconception/T1 vs T2/T3	1.01 (0.85–1.19)	1.04 (0.83–1.29)	1.06 (0.88–1.28)	0.98 (0.71–1.33)	0.99 (0.86–1.14)	0.86 (0.65–1.14)	1.24 (0.75–2.06)
T1 vs T2	1.11 (0.98–1.26)	1.10 (0.82–1.49)	1.17 (0.79–1.71)	1.03 (0.71–1.49)	1.19 (0.80–1.75)	0.88 (0.63–1.24)	1.53 (0.87–2.68)
T2 vs T3	0.78 (0.57–1.07)	9.33 (0.56–156.74)	0.88 (0.57–1.37)	8.66 (0.51–146.11)	0.71 (0.45–1.12)
T1 vs T3	0.94 (0.64–1.39)	15.85 (0.90–279.01)	1.32 (0.80–2.17)	10.97 (0.60–202.14)	1.10 (0.65–1.87)

Notes: Data are presented as RR (95% CI). Random-effects meta-analysis results for perinatal outcomes for different timings of ART initiation comparisons. A RR>1 indicates that the first-mentioned timing of ART initiation is associated with an increased risk of the corresponding perinatal outcome relative to the second-mentioned timing of ART initiation.
Abbreviations: CI, confidence interval; RR, relative risk; T1, first trimester; T2, second trimester; T3, third trimester.